Memory unit abnormity judgment method and server
After the memory stress test, the problem of high error judgment rate of memory unit abnormality determination in the prior art is solved, and the judgment is improved Accuracy and server stability.
Patent Information
- Application Number
- CN202411044764.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2024-11-26
AI Technical Summary
The existing memory unit exception determination method is prone to misjudgment of occasional CE as memory unit exception in memory stress tests, resulting in a high misjudgment rate and affecting the server's operating stability and reliability.
After the memory stress test is completed, the memory address to be tested is obtained and the memory address to be tested is determined based on the matching between the preset address set and the memory address to be tested. If it does not match, the memory stress test will be performed again, and the retest results will be obtained to determine the memory unit's Abnormal state, reducing the misjudgment rate.
It improves the accuracy of memory unit exception determination, reduces the misjudgment rate, and ensures the security and stability of the server operation.
Smart Images

Figure CN119025348A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of servers, and in particular to a method for determining abnormality of a memory unit and a server. Background Art
[0002] Memory stress testing is a test that evaluates the performance and stability of a computer or device under high memory usage conditions. Through memory stress testing, it can be determined whether the memory unit in the server is abnormal. During the memory stress test, due to the influence of the server operating environment, occasional CE may sometimes occur in the memory unit. The cause of this type of occasional CE often depends on the server's operating environment itself, rather than being caused by the abnormality of the memory unit itself. In the current memory unit abnormality determination method, occasional CE is prone to misjudgment of memory unit abnormality, and the misjudgment rate of memory unit abnormality determination is high. Summary of the invention
[0003] The embodiments of the present application provide a memory unit abnormality determination method and a server for improving the accuracy of abnormality determination when performing abnormality determination on a memory unit based on a memory stress test.
[0004] In a first aspect, an embodiment of the present application provides a method for determining an abnormality of a memory unit, the method comprising:
[0005] After the memory stress test is finished, obtaining a memory address to be tested of the memory unit to be tested; the memory unit to be tested is a memory unit in which a correctable error occurs during the memory stress test;
[0006] According to a preset address set and the memory address to be tested, determining whether the memory address to be tested matches the preset address set, and obtaining a matching result;
[0007] When the matching result is a mismatch, re-performing a memory stress test on the memory unit to be tested based on the memory stress test process to obtain a retest result;
[0008] When the correctable error exists in the retest result, it is determined that the memory state of the memory unit to be tested is an abnormal state.
[0009] In an embodiment of the present application, after the memory stress test is completed, the memory address to be tested of the memory unit to be tested where CE occurs is obtained. And based on whether the preset address set matches the memory address to be tested, it is determined whether the memory unit to be tested is abnormal. Since the memory addresses stored in the preset address set are memory addresses with a low probability of occasional CE, when the memory units to which these memory addresses belong generate CE, the probability that their CE is an occasional CE is low, so that the error rate of determining that the memory unit is abnormal is also reduced. At the same time, when the memory address to be tested does not match the address set, the memory stress test process to be tested is used to re-perform a memory stress test on the memory unit to be tested to obtain a corresponding retest result. And when the retest result has CE, it is determined that the memory unit to be tested is in an abnormal state, so as to determine whether the CE generated by the memory unit to be tested in the first memory test is an occasional CE, thereby further improving the accuracy of memory unit abnormality determination and reducing the error rate of memory unit abnormality determination.
[0010] In a possible implementation manner, after obtaining the memory address to be tested of the memory unit to be tested, the method further includes:
[0011] Obtaining the number of correctable errors generated by the memory unit to be tested during the memory stress test;
[0012] When the number of correctable errors is less than a preset correctable error number threshold, determining whether the memory address to be tested matches the preset address set to obtain the matching result; the preset correctable error number threshold is determined based on the memory test unit to be tested;
[0013] When the matching result is a match, it is determined that the memory state of the memory unit to be tested is the abnormal state.
[0014] In this implementation, after the memory stress test for multiple memory units is completed, for the memory unit to be tested in which CE appears, before judging whether the memory address to be tested matches the preset address set, it is necessary to make a preliminary abnormal judgment on the state of the memory unit to be tested based on the total number of CEs generated in the memory unit to be tested. If the number of CEs generated by the memory unit to be tested is less than the preset threshold value of the number of correctable errors, the above matching judgment between the memory address to be tested and the preset address set is performed to obtain a corresponding matching result. The memory address stored in the preset address set is used to represent the memory area of the operating system and the core application in the server device, which is a key part of the operation of the server device. When CE appears in the memory area of these key parts, it may have a greater impact on the server device, thereby reducing the operation stability and reliability of the server device. Therefore, in order to prevent such a situation from happening, when the memory units of these key parts generate CE after the memory stress test, that is, when the matching result obtained is a match, it is necessary to directly determine that there is an abnormality in its memory unit, so as to prevent potential operation crashes or data damage of the server device, so as to ensure the operation safety of the server device.
[0015] In a possible implementation manner, different memory units to be tested correspond to different preset correctable error number thresholds; and a method for determining the preset correctable error number threshold specifically includes:
[0016] Obtaining unit attribute information and a historical correctable error log of the memory unit to be tested; the unit attribute information includes: location information, usage time information, and configuration information of the memory unit to be tested; the historical correctable error log is used to indicate the number of correctable errors that occurred in the memory unit to be tested within a preset historical time period;
[0017] Cluster analysis is performed based on the unit attribute information of the memory unit to be tested and the historical correctable error log to determine the preset correctable error quantity threshold corresponding to the memory unit to be tested.
[0018] In this implementation, the storage specification and historical performance data of the memory unit to be tested are used as reference factors for setting the corresponding preset correctable error number threshold, which can effectively determine the preset correctable error number threshold of the memory unit to be tested. When the number of CEs in the memory unit to be tested is not less than the preset correctable error number threshold, it is directly determined that there is an abnormality in the memory stress test of the memory unit to be tested, without the need to perform the subsequent matching judgment steps between the memory address to be tested and the preset address set, so that in the initial stage of abnormal judgment of the memory unit to be tested, a preliminary judgment is made based on the number of CEs in the memory unit to be tested, which effectively improves the efficiency of abnormal judgment of the memory unit.
[0019] In a possible implementation manner, after obtaining the number of correctable errors generated by the memory unit to be tested in the memory stress test, the method further includes:
[0020] When the number of correctable errors is not less than the preset correctable error number threshold, it is determined that the memory state of the memory unit to be tested is the abnormal state.
[0021] In the present implementation, when the number of correctable errors generated by the memory unit to be tested and when the number of CEs in the memory unit to be tested is not less than a preset correctable error number threshold, it is directly determined that there is an abnormality in the memory stress test of the memory unit to be tested, and there is no need to perform subsequent steps of matching and judging between the memory address to be tested and the preset address set. Therefore, in the initial stage of abnormal judgment of the memory unit to be tested, a preliminary judgment is made based on the number of CEs in the memory unit to be tested, which effectively improves the efficiency of abnormal judgment of the memory unit.
[0022] In a possible implementation, obtaining the memory address to be tested of the memory unit to be tested includes:
[0023] Obtain the type of stress test tool for performing the memory stress test;
[0024] According to the stress testing tool type, determining a memory stress testing log corresponding to the stress testing tool type;
[0025] Perform correctable error analysis on the memory stress test log to obtain the memory address to be tested.
[0026] In this implementation, by obtaining the type of test tool used when performing memory stress testing, the memory stress test log of the server can be accurately obtained. By analyzing the correctable errors of the memory stress test log, the memory address to be tested can be accurately obtained, thereby improving the efficiency of obtaining the memory address to be tested.
[0027] In a possible implementation, after obtaining the retest result, the method further includes:
[0028] When the correctable error does not exist in the retest result, it is determined that the memory state of the memory unit to be tested is a normal state.
[0029] In this implementation, when there is no CE in the retest results of the memory unit to be tested, it indicates that the CE generated by the memory unit to be tested in the first memory stress test is an occasional CE, and it can be determined that the memory state of the memory unit to be tested is normal, which reduces the misjudgment rate of the memory unit abnormality judgment and improves the accuracy of the memory unit abnormality judgment.
[0030] In a possible implementation, before obtaining the memory address of the memory unit to be tested, the method further includes:
[0031] Obtain the test task status of the memory stress test;
[0032] When the test task status is the execution end status, the step of obtaining the memory address to be tested of the memory unit to be tested is performed.
[0033] In this implementation, by obtaining the test task status of the memory stress test, when the test task status is the execution end state, the step of obtaining the memory address to be tested of the memory unit to be tested is executed. After the memory stress test on multiple memory units is completed, the memory units to be tested that have CE are uniformly judged as abnormal, which can improve the efficiency of abnormal judgment of memory units.
[0034] In a possible implementation, obtaining the test task status of the memory stress test includes:
[0035] When the execution of a single memory stress test algorithm ends or the execution of all the memory stress test algorithms ends, determining that the test task state is the execution end state;
[0036] When the memory pressure test algorithm is not finished executing, the test task state is determined to be an executing state.
[0037] In this implementation, when no memory stress test algorithm has been executed, the task state of the memory stress test is determined to be in the executing state. When a single memory stress test algorithm has been executed or all memory stress test algorithms have been executed, the test task state of the memory stress test is determined to be in the execution end state, so that the abnormality determination of the memory unit can be performed after at least one memory stress test algorithm has been executed, thereby improving the abnormality determination efficiency of the memory unit.
[0038] In a possible implementation, the memory test process to be tested is obtained based on a mapping relationship between the memory address to be tested and a preset address; the preset address mapping relationship is used to represent a mapping relationship between a memory address and a memory stress test process.
[0039] In this implementation, the memory test process to be tested can be determined based on the memory address to be tested and the preset address mapping relationship. The preset address mapping relationship includes a mapping relationship between multiple memory addresses and corresponding test processes. By using the memory address to be tested as an index, the memory test process to be tested corresponding to the memory address to be tested in the preset address mapping relationship can be obtained, thereby ensuring the effective determination of the memory test process to be tested.
[0040] In a second aspect, an embodiment of the present application provides a memory unit abnormality determination device, the device comprising:
[0041] An acquisition unit, used for acquiring a memory address to be tested of a memory unit to be tested after the memory stress test is completed; the memory unit to be tested is a memory unit in which a correctable error occurs during the memory stress test;
[0042] A first determining unit, configured to determine, based on a preset address set and the memory address to be tested, whether the memory address to be tested matches the preset address set, and obtain a matching result;
[0043] A retest unit, configured to, when the matching result is a mismatch, re-perform a memory stress test on the memory unit to be tested based on the memory stress test process to obtain a retest result;
[0044] The second determining unit is used to determine that the memory state of the memory unit to be tested is an abnormal state when the correctable error exists in the retest result.
[0045] In a third aspect, an embodiment of the present application provides a server, comprising: a processor, and a memory communicatively connected to the processor;
[0046] The memory is used to store computer-executable instructions;
[0047] The processor is used to execute the computer-executable instructions stored in the memory to implement the memory unit abnormality determination method described in any embodiment / implementation of any of the aforementioned aspects.
[0048] In a fourth aspect, an embodiment of the present application provides a computer program product, which, when executed on a computer, executes the operating steps of any possible memory unit abnormality determination method of the first aspect.
[0049] Any of the memory unit abnormality determination methods, memory unit abnormality determination devices, servers or computer program products provided above are used to execute the corresponding methods provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding methods and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] In order to more clearly illustrate the technical solutions in this embodiment or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0051] Figure 1 A schematic diagram of a method for determining a memory unit abnormality in an actual application scenario provided by an embodiment of the present application;
[0052] Figure 2A schematic diagram of a flow chart of a method for determining abnormality of a memory unit provided in an embodiment of the present application;
[0053] Figure 3 A schematic diagram of a flow chart of another memory unit abnormality determination method provided in an embodiment of the present application;
[0054] Figure 4 A schematic diagram of a mapping relationship between a test process identifier and a memory address disclosed in an embodiment of the present application;
[0055] Figure 5 A flowchart of another memory unit abnormality determination method provided in an embodiment of the present application;
[0056] Figure 6 A flowchart of a method for determining a preset threshold value of the number of correctable errors provided in an embodiment of the present application;
[0057] Figure 7 A schematic diagram of another memory unit abnormality determination method provided in an embodiment of the present application;
[0058] Figure 8 A schematic diagram of another memory unit abnormality determination method provided in an embodiment of the present application. DETAILED DESCRIPTION
[0059] It should be noted that the embodiments described in this application are only part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.
[0060] In order to make the following embodiments clear, the technical terms involved in this application are first introduced.
[0061] The operating system (OS) is one of the core software in a computer system. It is responsible for managing computer hardware resources and providing operating interfaces and services for users and applications. Common operating systems include Windows, macOS, Linux, and Android.
[0062] Correctable Error (CE), in computer systems, some errors may occur in memory modules, such as bit flipping, noise interference, or other factors that cause data corruption. Some of these errors are correctable, which means that the system is able to detect the errors and automatically repair them through correction codes or other mechanisms without user intervention. This type of error is often called a correctable error, and CE may refer to these correctable memory errors.
[0063] The memory unit abnormality determination method provided in the embodiment of the present application will be described in detail below with reference to the accompanying drawings of specific embodiments. Figure 1 , this figure is a schematic diagram of a memory unit abnormality determination method in an actual application scenario provided by an embodiment of the present application. In the actual application scenario, the server 101 may be a server for whole machine production testing.
[0064] like Figure 1 As shown, when the server 101 receives the memory stress test instruction, it performs a memory stress test on multiple memory units in the memory module according to the memory stress test instruction. The memory stress test instruction can be generated by a test tool, such as Memtester, Prime95, etc. These test tools can run in the operating system in the server according to a specific algorithm and perform a memory stress test on the memory module of the server.
[0065] In the process of the server performing memory stress testing on multiple memory units in the memory module, different memory units correspond to different types of memory stress testing algorithms, and different memory stress testing algorithms correspond to different test processes. Among them, the type of memory stress testing algorithm can be a variety of test algorithm types such as random access test, sequential access test, pattern test, stress test, durability and life test, and voltage boundary test. The type of test algorithm is determined by the type of memory unit and the use environment of the server device. This embodiment does not specifically limit the type of memory stress testing algorithm.
[0066] After the memory stress test for multiple memory units is completed, the memory unit where CE appears is determined as the memory unit to be tested from the memory module, and it is determined whether the memory address to be tested where the memory unit to be tested appears is a memory address in the preset address set. Among them, the memory addresses stored in the preset address set are memory addresses with a low probability of occasional CE. The memory addresses stored in the preset address set are the memory addresses that have the most critical impact on the operation of the server. In actual application scenarios, the memory addresses that have the most critical impact on the operation of the server are often the memory addresses of the operating system or the memory addresses of the core application in the server device. Since these memory addresses are the memory addresses that have the most critical impact on the operation of the server, the operating environment of the memory units corresponding to these addresses are strictly controlled, so the probability of them generating CE is very low.
[0067] If the memory address to be tested does not match the address in the preset address set, it indicates that there is a certain probability of sporadic CE at the memory address to be tested. Therefore, when the memory address to be tested does not match the address in the preset address set, based on the memory stress test process to be tested corresponding to the memory address to be tested, the memory unit to be tested is re-tested with the memory stress test process to obtain the corresponding retest result. If there is still CE in the retest result, it indicates that the CE generated by the memory unit to be tested during the first memory stress test is not a sporadic CE, and at this time, it can be determined that the memory state of the memory unit to be tested is an abnormal state.
[0068] In this way, by dividing the preset address set, the memory unit with a higher probability of occasional CE can be divided separately, and the memory stress test can be re-performed on it, and the retest result can be used to determine whether the memory unit to be tested is actually abnormal. Since the CE with a lower probability of occasional CE has been separated by the preset address set, this method can greatly reduce the problem of high misjudgment rate of memory unit abnormality judgment caused by occasional CE, and improve the accuracy of memory unit abnormality judgment.
[0069] On the other hand, since the memory addresses in the preset address set are the memory addresses of the operating system and the core application, they are the most critical memory addresses during the operation of the server device. Therefore, when CE appears in the memory unit that matches the preset address set, the memory unit with CE will have a greater impact on the operation of the server device, thereby reducing the operational stability and reliability of the server device. Therefore, after the memory stress test of multiple memory units is completed, when the memory address with CE matches the address in the preset address set, it is necessary to directly determine that the memory stress test process corresponding to the memory address has an abnormality, thereby ensuring the operational safety of the server.
[0070] At the same time, since the judgment on the memory address to be tested and the preset address set is performed only after the memory stress test is completed, even if CE appears in the memory unit during the memory stress test, the memory stress test will not be interrupted. Only after the memory stress test of all memory units is completed will the abnormality judgment on the memory unit be performed, thereby improving the judgment efficiency of the memory unit abnormality judgment.
[0071] Among them, the memory address of the memory unit can be determined by the test tool used to perform memory stress testing. Taking the above-mentioned Memtester tool as an example, in the process of using the Memtester tool to perform memory stress testing, when a CE appears in a memory unit, the time when the unit generates CE and the address of the corresponding memory unit will be saved in the mceaddr file. Mceaddr is a file automatically generated when using the Memtester tool for memory stress testing. It is used to record the specific time when CE appears in the memory unit and the corresponding memory address. When determining the memory address of the memory unit, after the memory stress test is completed, by reading the mceaddr file in the Memtester tool, the memory address of the memory unit can be successfully obtained, and the matching judgment between the memory address to be tested and the preset address set can be performed.
[0072] Next, a memory unit abnormality determination method provided in an embodiment of the present application will be introduced in conjunction with the accompanying drawings.
[0073] See also Figure 2 , which is a flow chart of a method for determining abnormality of a memory unit provided in an embodiment of the present application, and specifically includes the following steps:
[0074] S101: After the memory stress test is finished, obtaining a memory address of a memory unit to be tested; the memory unit to be tested is a memory unit in which a correctable error occurs during the memory stress test.
[0075] The memory unit abnormality determination method of the embodiment of the present application is applied to a server device, in which a memory module includes multiple memory units. Since different memory units are used to store different types of information and log files, the memory stress tests corresponding to different memory units are also different.
[0076] Memory stress testing needs to divide multiple different test processes based on the type of memory unit, and different test processes correspond to different test algorithms. In actual application scenarios, there may be multiple memory units in a server device corresponding to the same memory stress testing process, that is, multiple memory units use the same test algorithm to perform memory stress testing. In the memory unit abnormality determination method provided in the embodiment of the present application, since one memory stress testing process corresponds to multiple memory units, the end of the memory stress test can be regarded as the end of the execution of one memory stress testing algorithm, or it can be regarded as the end of the execution of all memory stress testing algorithms.
[0077] Therefore, the step of obtaining the memory address to be tested can be performed after a single memory stress test algorithm is executed, or after all memory stress test algorithms are executed. In view of the above two different situations, the abnormal judgment of the test process is carried out after the memory stress test under the execution of a single memory test algorithm is completed. This method has higher real-time performance and can provide faster feedback on the performance and stability of a single process, which helps to quickly identify and solve problems. Correspondingly, if the abnormal judgment is carried out after all memory test algorithms are executed, this method can indirectly reflect the comprehensive performance capabilities of the server under high-pressure conditions, and is also suitable for scenarios that evaluate the entire server.
[0078] It should be noted that although the two abnormality determination timings are different in specific operations, the essence of both is to uniformly perform abnormality determination on the memory cells that generate CE after multiple memory stress test processes are completed. This determination method can avoid the problem of low efficiency of abnormality determination of memory cells caused by the previous abnormality determination method of memory cells, which immediately determines the abnormality of the memory cells after CE appears in the memory cells, thereby improving the abnormality determination efficiency for memory cells.
[0079] Next, the determination link of whether the memory stress test is finished before obtaining the memory address to be tested will be introduced in conjunction with the drawings of the specific embodiment.
[0080] See also Figure 3 , which is a flow chart of another memory unit abnormality determination method provided in an embodiment of the present application, and specifically includes the following steps:
[0081] S401: Obtain the test task status of the memory stress test.
[0082] When the server performs memory stress testing on multiple memory units, it is necessary to obtain the test task status of the memory stress test in real time. From the above description, it can be seen that in order to improve the abnormality determination efficiency of the memory unit, the process of obtaining the memory address to be tested needs to be carried out after a single memory stress test algorithm or all memory stress test algorithms are completed.
[0083] Therefore, the test task status is used to indicate whether a memory stress test algorithm has ended in the current memory stress test of the server. When a single memory stress test algorithm has ended or all memory stress test algorithms have ended, the test task status is an execution end state. When no memory stress test algorithm has ended, the test task status is an executing state. Among them, the test task status of the memory stress test is stored in the test log generated in real time by the test tool used by the server to perform memory stress testing. By reading the test log generated in real time, the test task status of the memory stress test can be obtained in real time.
[0084] S402: When the test task status is an execution end status, the step of obtaining the memory address to be tested of the memory unit to be tested is executed.
[0085] When it is determined that the test task status of the memory stress test is the execution end status, it indicates that the execution of the memory stress test algorithm in the memory stress test has ended, and the memory unit that is stress tested in a memory stress test algorithm has been tested. At this time, for multiple memory units that have passed the memory stress test, the step of obtaining the memory address to be tested is performed on the memory unit to be tested that generates CE, which can effectively improve the efficiency of abnormal determination of memory units.
[0086] When the test task status of the memory stress test is in the executing state, it indicates that no memory stress test algorithm has been completed in the server at this time, and the step of obtaining the memory address to be tested will not be executed, so that the abnormality judgment of the memory unit can be performed after at least one memory stress test algorithm is completed, thereby improving the abnormality judgment efficiency of the memory unit.
[0087] Specifically, by Figure 1 As can be seen from the description of the embodiment, the memory stress test for multiple memory units in the server can be completed by different memory stress test tools. And for different memory stress test tools, the corresponding log files also have certain differences. Therefore. In the actual application scenario, when determining the memory address of the memory unit to be tested, it needs to be completed according to the type of stress test tool, and its specific process is completed through the following three steps:
[0088] Step 1: Obtain the type of stress test tool for performing the memory stress test.
[0089] In the process of determining the memory address to be tested of the memory unit to be tested, it is necessary to first determine the tool type of the memory stress test tool. Commonly used memory stress test tools in the prior art include Memtester, Prime95, and Stress-ng.
[0090] Step 2: According to the stress testing tool type, determine the memory stress testing log corresponding to the stress testing tool type.
[0091] Because the log formats and storage paths generated by different memory stress testing tools may be different, it is necessary to determine the storage path and storage format of the memory stress testing logs based on the specific test tool type in order to read the required logs.
[0092] In actual application scenarios, a path mapping table can be pre-established to map the tool type of the memory stress test tool to the corresponding log storage path, and the log file in the corresponding format can be parsed based on the type of the test tool to obtain the memory stress test log.
[0093] Step three: performing correctable error analysis on the memory stress test log to obtain the memory address to be tested.
[0094] Further, after obtaining the memory stress test log, the log file is subjected to correctable error analysis, thereby identifying the location where CE appears in the log. By reading the log at this location, the memory address to be tested of the memory unit to be tested can be obtained.
[0095] In the process of correctable error analysis of the test log, the keyword search method can be used to determine the location of CE in the log. Common keywords include "CE", "correctable error", "memory error", etc. After finding the keyword, perform context analysis on the log at the keyword to understand the time when CE occurred and related information, and then determine the required memory address to be tested.
[0096] S102: According to a preset address set and the memory address to be tested, determining whether the memory address to be tested matches the preset address set, and obtaining a matching result.
[0097] After determining the memory address to be tested where CE appears, further match and compare the memory address to be tested with the preset address set to obtain a matching result. And based on the matching result, determine whether the memory address in the preset address set is the memory address to be tested where CE appears. Since the addresses in the preset address set are the memory addresses that have the most critical impact on the operation of the server, these memory addresses are usually stable operating areas in the server. When CE appears in these memory addresses, the probability that the CE is an occasional CE is low. Therefore, when the memory stress test process corresponding to these memory addresses generates CE, it can be directly determined that the memory stress test process is abnormal, thereby reducing the misjudgment rate of the memory stress test.
[0098] Among them, for the memory addresses in the preset address set, the probability of occasional CE occurring in the memory addresses can be described by a preset first threshold. By analyzing the historical operation data of the memory units to which each memory address in the preset address set belongs, and counting the frequency of CE occurring in each memory unit in the memory stress test, the preset first threshold can be set according to the obtained statistical results, thereby characterizing the probability of occasional CE occurring in the memory addresses in the preset address set.
[0099] S103: When the matching result is a mismatch, re-perform a memory stress test on the memory unit to be tested based on the memory stress test process to obtain a retest result.
[0100] Correspondingly, when the matching result is mismatched, it indicates that the memory address to be tested does not match the preset address set, and the probability of occasional CE in the memory unit to be tested to which the memory address to be tested belongs is relatively high. Therefore, in order to prevent occasional CE from having a negative impact on the accuracy of abnormal determination of memory units, when the memory address to be tested does not match the preset address set, it is necessary to re-perform a memory stress test on the memory unit to be tested through the memory stress test process to obtain the corresponding retest result, so as to judge whether the memory state of the memory unit to be tested is abnormal according to whether the memory unit to be tested generates CE in the retest result.
[0101] The memory test process to be tested may be the test process used by the memory unit to be tested in the first memory stress test, or may be a reselected test process that is stricter than the test process used in the first memory stress test.
[0102] In particular, when the memory test process to be tested uses the test process used in the first memory stress test, the memory test process to be tested can be determined by the preset address mapping relationship between each memory address and the memory stress test process. As can be seen from the foregoing, the memory addresses of different memory units correspond to different test processes, such as Figure 4 As shown in the figure, pid1-pidx represent different test process identifiers, and each test process identifier corresponds to a memory address of a certain length. In actual application scenarios, there are often multiple memory units in a memory address, and the multiple memory units are of the same type and correspond to a single test process identifier. At the end of the memory stress test, the memory address to be tested where CE appears is determined, and based on the mapping relationship between the pre-set test process identifier and the memory address, the memory stress test process to be tested corresponding to the memory address to be tested can be determined.
[0103] It should be noted that when the memory stress test is performed for the first time, the memory stress test is performed on multiple memory units, and the memory stress test includes multiple test processes. However, in the process of re-testing the memory unit to be tested through the memory stress test process to be tested, the server only runs the memory stress test process to be tested for the memory unit to be tested, and the memory unit that did not generate CE during the first memory stress test will not be re-tested, so as to improve the efficiency of determining abnormalities in memory units.
[0104] S104: When the correctable error exists in the retest result, determining that the memory state of the memory unit to be tested is an abnormal state.
[0105] When there is a CE in the retest result of the memory unit to be tested, it indicates that the CE generated by the memory unit to be tested during the first memory stress test is not an occasional CE, and at this time, it can be determined that the memory state of the memory unit to be tested is an abnormal state.
[0106] In this way, by dividing the preset address set, the memory unit with a higher probability of occasional CE can be divided separately, and the memory stress test can be re-performed on it, and the retest result can be used to determine whether the memory unit to be tested is actually abnormal. Since the CE with a lower probability of occasional CE has been separated by the preset address set, this method can greatly reduce the problem of high misjudgment rate of memory unit abnormality judgment caused by occasional CE, and improve the accuracy of memory unit abnormality judgment.
[0107] The following will further explain the steps after obtaining the address of the memory unit to be tested in combination with the specific embodiment drawings:
[0108] See also Figure 5 , which is a flow chart of another memory unit abnormality determination method provided in an embodiment of the present application, and specifically includes the following steps:
[0109] S201: Obtain the number of correctable errors generated by the memory unit to be tested in the memory stress test.
[0110] After the memory stress test for multiple memory units is completed, for the memory unit to be tested in which CE appears, before judging whether the memory address to be tested matches the preset address set, it is necessary to make a preliminary abnormality judgment on the state of the memory unit to be tested based on the total number of CEs generated in the memory unit to be tested.
[0111] Among them, the number of CEs generated by the memory unit to be tested in the memory stress test can be obtained through the test log in the test tool used to perform the memory stress test. In a conventional memory stress test tool, the number of CEs generated by each memory unit in the memory stress test will be displayed in its test log, so that the number of CEs corresponding to the memory unit to be tested can be obtained.
[0112] S202: When the number of correctable errors is less than a preset correctable error number threshold, determine whether the memory address to be tested matches the preset address set to obtain the matching result; the preset correctable error number threshold is determined based on the memory stress test process to be tested.
[0113] In the process of making a preliminary abnormal judgment on the memory unit to be tested, the relationship between the number of CEs generated by it and the preset threshold of the number of correctable errors is needed to make an abnormal judgment. If the number of CEs generated by the memory unit to be tested is not less than the preset threshold of the number of correctable errors, it means that the number of CEs generated by the memory unit to be tested in the memory stress test is too much, exceeding the maximum number of occurrences of sporadic CEs itself, and thus it can be directly determined that the memory state of the memory unit to be tested is abnormal. Correspondingly, if the number of CEs generated by the memory unit to be tested is less than the preset threshold of the number of correctable errors, the above-mentioned matching judgment between the memory address to be tested and the preset address set is executed to obtain the corresponding matching result.
[0114] Furthermore, the setting of the preset correctable error number threshold is determined based on the cell attribute information of the memory cell to be tested, and each memory cell to be tested may have a different preset correctable error number threshold due to its own different cell attribute conditions.
[0115] S203: When the matching result is a match, determining that the memory state of the memory unit to be tested is the abnormal state.
[0116] As can be seen from the previous description, the memory addresses stored in the preset address set are used to represent the memory areas of the operating system and core applications in the server device, which are key parts of the server device operation. When CE appears in the memory areas of these key parts, it may have a greater impact on the server device, thereby reducing the operational stability and reliability of the server device. Therefore, in order to prevent such situations from occurring, when the memory units of these key parts generate CE after the memory stress test, that is, when the matching result obtained is a match, it is necessary to directly determine that there is an abnormality in its memory unit, thereby preventing potential server device operation crashes or data corruption, to ensure the safe operation of the server device.
[0117] Specifically, the process of determining the preset correctable error number threshold value through the cell attribute information of the memory cell to be tested and the historical correctable error log is completed in the following manner:
[0118] See also Figure 6 , which is a flow chart of a method for determining a preset threshold value of the number of correctable errors provided in an embodiment of the present application, and specifically includes the following steps:
[0119] S301: Obtain unit attribute information and historical correctable error log of the memory unit to be tested; the unit attribute information includes: location information, usage time information and configuration information of the memory unit to be tested; the historical correctable error log is used to indicate the number of correctable errors that occurred in the memory unit to be tested within a preset historical time period.
[0120] It is understandable that in actual application scenarios, different memory units have different physical locations, workloads, hardware configurations, and total usage time. And as the usage time of the memory unit increases, its internal hardware may age. Therefore, for the determination of the preset correctable error number threshold for the memory unit to be tested, it is necessary to combine the unit attribute information of the memory unit to be tested itself, as well as the situation of CE occurring in the memory unit to be tested within the historical time to set it. Among them, the situation of CE occurring in the memory unit to be tested within the historical time is represented by the historical correctable error log. Through the historical correctable error log, the number of correctable errors generated by the memory unit to be tested within a certain preset historical time can be determined, thereby characterizing the hardware situation of the memory unit to be tested itself.
[0121] Furthermore, the unit attribute information of the memory unit to be tested may also be storage specifications and historical performance data. The storage specifications indicate the data capacity that the memory unit to be tested can store, and the historical performance data may be the frequency and total number of CEs that appear in the memory unit to be tested in historical time. The storage specifications and historical performance data of the memory unit to be tested are used as reference factors for setting the corresponding preset correctable error number threshold, which can effectively determine the preset correctable error number threshold of the memory unit to be tested. When the number of CEs in the memory unit to be tested is not less than the preset correctable error number threshold, it is directly determined that there is an abnormality in the memory stress test of the memory unit to be tested, without the need to perform the subsequent matching judgment steps between the memory address to be tested and the preset address set, so that in the initial stage of abnormal judgment of the memory unit to be tested, a preliminary judgment is made based on the number of CEs in the memory unit to be tested, which effectively improves the efficiency of abnormal judgment of the memory unit.
[0122] S302: Perform cluster analysis based on the unit attribute information of the memory unit to be tested and the historical correctable error log to determine the preset correctable error quantity threshold corresponding to the memory unit to be tested.
[0123] The preset threshold value of the number of correctable errors is determined by clustering analysis of the unit attribute information and the historical correctable error log. In the process of clustering analysis, it is necessary to first extract features of the information with a greater degree of influence in the unit attribute information, such as the physical location of the memory unit (slot location or module location), manufacturing batch (manufacturing date, manufacturer, batch number), workload type (application type, load intensity, access mode), environmental factors (temperature, humidity, power supply stability) and hardware configuration and other feature information, extract features from various types of information, and combine the number of correctable errors that appear in the historical correctable error log of the memory unit to obtain features for clustering analysis.
[0124] After that, according to a specific clustering algorithm, such as K-means and hierarchical clustering, the obtained features are clustered and analyzed to obtain a corresponding cluster group. By performing a statistical analysis on the number of correctable errors in the cluster group, the mean, median, standard deviation and other data are calculated. The correctable error number threshold is set based on the statistical analysis results, and all memory cells can be traversed in this way to determine the specific preset correctable error number threshold for each memory cell.
[0125] In order to further understand the technical solution of the embodiment of the present application, a memory unit abnormality determination method provided by the embodiment of the present application is introduced in detail below in combination with actual application scenarios.
[0126] See also Figure 7 , which is a schematic diagram of another memory unit abnormality determination method provided in an embodiment of the present application, and it should be noted in advance that, Figure 7 The memory unit abnormality determination method shown is applied to the situation where each memory unit corresponds to the same memory test algorithm 1. From the foregoing, it can be seen that the memory unit abnormality determination method of the present application can perform abnormality determination after the execution of a single memory stress test algorithm is completed, or it can perform abnormality determination after the execution of all memory stress test algorithms is completed. There are certain differences in the implementation effects of the two, but the essence is that after the memory stress test of multiple memory units is completed, abnormality determination is uniformly performed on the memory units that generate CE. The two different processing forms have the same principle, and both can improve the efficiency of memory unit abnormality determination. Therefore, Figure 7 The example is only an optional implementation, and does not limit the abnormality determination logic between the memory test algorithm and the memory unit.
[0127] See also Figure 8 , which is a schematic diagram of another memory unit abnormality determination method provided in an embodiment of the present application, which can be specifically combined with Figure 8 and Figure 7 To further understand. Specifically, Figure 7 The schematic diagram includes the following steps:
[0128] S1101: Determine whether CE appears in the memory unit;
[0129] After all memory unit tests in the memory test algorithm 1 are completed, it is necessary to obtain the corresponding memory address to be tested according to the memory unit to be tested in which CE occurs. Therefore, after the memory test algorithm 1 is completed, it is necessary to determine whether the memory unit in the memory test algorithm 1 has CE.
[0130] S1102: Determine that the memory status of all memory units of memory test algorithm 1 is normal.
[0131] When no CE occurs in multiple memory cells in memory test algorithm 1, it indicates that the memory status of the memory cells in memory test algorithm 1 is normal. At this time, memory test algorithm 2 will be directly entered to carry out memory stress test for other memory cells through memory test algorithm 2.
[0132] S1103: Determine whether the number of correctable errors generated by the memory unit to be tested is greater than a preset correctable error threshold.
[0133] When a memory unit to be tested has CE, it is necessary to obtain the number of correctable errors generated by the memory unit to be tested in the initial memory test algorithm 1, and determine whether the number of CEs is greater than the preset threshold value of the number of correctable errors. In this way, a preliminary abnormality determination can be made on the memory unit to be tested based on the number of CEs generated by the memory unit to be tested in the memory test algorithm 1, and when the number of CEs is not less than the preset threshold value of correctable errors, the memory state of the memory unit to be tested is directly determined to be an abnormal state, thereby improving the efficiency of abnormality determination.
[0134] S1104: Determine whether the memory address to be tested of the memory unit to be tested is an address in a preset address set.
[0135] Then, when the number of CEs generated by the memory unit to be tested is less than the preset threshold value of the number of correctable errors, it is determined whether the memory address to be tested is an address in the preset address set. Since the addresses in the preset address set are the memory addresses that have the most critical impact on the operation of the server, these memory addresses are usually stable operation areas in the server. When CEs appear in these memory addresses, the probability that the CE is an occasional CE is low. Therefore, when the memory units corresponding to these memory addresses generate CEs, it can be directly determined that the memory unit is abnormal, thereby reducing the misjudgment rate of memory unit abnormality determination.
[0136] S1105: re-stress-test the memory unit to be tested through the memory stress test process to obtain a re-test result.
[0137] When the memory address to be tested does not match the preset address set, in order to prevent occasional CE from affecting the accuracy of memory unit abnormality judgment, it is necessary to re-perform memory stress test on the memory unit to be tested based on the memory stress test process to obtain the corresponding retest result.
[0138] S1106: Determine whether a correctable error occurs in the retest result.
[0139] Subsequently, it is determined whether CE appears in the retest result of the memory unit to be tested. When CE exists in the retest result of the memory unit to be tested, it indicates that the CE generated by the memory unit to be tested in the memory test algorithm 1 is not an occasional CE. At this time, it can be determined that the memory state of the memory unit to be tested is an abnormal state.
[0140] S1107: Determine that the memory state of the memory unit to be tested is normal.
[0141] Correspondingly, when there is no CE in the retest result, it indicates that the CE generated by the memory unit to be tested in the memory test algorithm 1 is an occasional CE, and the memory state of the memory unit itself is normal. At this time, it can be determined that the memory state of the memory unit to be tested is normal.
[0142] S1108: Determine that the memory state of the memory unit to be tested is an abnormal state.
[0143] The embodiment of the present application also provides a memory unit abnormality determination device, the device comprising:
[0144] An acquisition unit, used for acquiring a memory address to be tested of a memory unit to be tested after the memory stress test is completed; the memory unit to be tested is a memory unit in which a correctable error occurs during the memory stress test;
[0145] A first determining unit, configured to determine, based on a preset address set and the memory address to be tested, whether the memory address to be tested matches the preset address set, and obtain a matching result;
[0146] A retest unit, configured to, when the matching result is a mismatch, re-perform a memory stress test on the memory unit to be tested based on the memory stress test process to obtain a retest result;
[0147] The second determining unit is used to determine that the memory state of the memory unit to be tested is an abnormal state when the correctable error exists in the retest result.
[0148] The embodiment of the present application also provides a server, comprising: a processor, and a memory communicatively connected to the processor;
[0149] The memory is used to store computer-executable instructions;
[0150] The processor is used to execute the computer-executable instructions stored in the memory to implement the memory unit abnormality determination method of any embodiment / implementation method of the present application.
[0151] An embodiment of the present application provides a computer program product, which, when executed on a computer, executes the operating steps of the memory unit abnormality determination method of any embodiment / implementation of the present application.
[0152] The "first" and "second" in the names such as "first" and "second" (if any) mentioned in the embodiments of the present application are only used as name identifiers and do not represent the first or second in order.
[0153] Through the description of the above implementation methods, it can be known that those skilled in the art can clearly understand that all or part of the steps in the above-mentioned embodiment method can be implemented by means of software plus a general hardware platform. Based on such an understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a storage medium, such as a read-only memory (ROM) / RAM, a magnetic disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network communication device such as a router) to execute the methods described in each embodiment of the present application or some parts of the embodiments.
[0154] Each embodiment in this specification is described in a progressive manner, and the same and similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative work.
[0155] The above description is merely an exemplary embodiment of the present application and is not intended to limit the protection scope of the present application.
Claims
1. A method for determining abnormality of a memory unit, characterized in that: include: After the memory stress test is completed, the memory address to be tested of the memory unit to be tested is obtained; The memory unit to be tested is a memory unit in which a correctable error occurs during the memory stress test; According to the preset address set and the memory address to be tested, determining whether the memory address to be tested matches the preset address set, and obtaining a matching result; When the matching result is a mismatch, re-performing a memory stress test on the memory unit to be tested based on the memory stress test process to obtain a retest result; When the correctable error exists in the retest result, it is determined that the memory state of the memory unit to be tested is an abnormal state.
2. The method according to claim 1, characterized in that After obtaining the memory address of the memory unit to be tested, the method further includes: Obtaining the number of correctable errors generated by the memory unit to be tested during the memory stress test; When the number of correctable errors is less than a preset correctable error number threshold, determining whether the memory address to be tested matches the preset address set to obtain the matching result; the preset correctable error number threshold is determined based on the memory test unit to be tested; When the matching result is a match, it is determined that the memory state of the memory unit to be tested is the abnormal state.
3. The method according to claim 2, characterized in that Different memory units to be tested correspond to different preset correctable error quantity thresholds; The method for determining the preset correctable error quantity threshold specifically includes: Obtaining unit attribute information and a historical correctable error log of the memory unit to be tested; the unit attribute information includes: location information, usage time information, and configuration information of the memory unit to be tested; the historical correctable error log is used to indicate the number of correctable errors that occurred in the memory unit to be tested within a preset historical time period; Cluster analysis is performed based on the unit attribute information of the memory unit to be tested and the historical correctable error log to determine the preset correctable error quantity threshold corresponding to the memory unit to be tested.
4. The method according to claim 2, characterized in that: After obtaining the number of correctable errors generated by the memory unit to be tested in the memory stress test, the method further includes: When the number of correctable errors is not less than the preset correctable error number threshold, it is determined that the memory state of the memory unit to be tested is the abnormal state.
5. The method according to claim 1, characterized in that The step of obtaining the memory address to be tested of the memory unit to be tested comprises: Obtain the type of stress test tool for performing the memory stress test; According to the stress testing tool type, determining a memory stress testing log corresponding to the stress testing tool type; Perform correctable error analysis on the memory stress test log to obtain the memory address to be tested.
6. The method according to claim 1, characterized in that After obtaining the retest results, the method further comprises: When the correctable error does not exist in the retest result, it is determined that the memory state of the memory unit to be tested is a normal state.
7. The method according to claim 1, characterized in that Before obtaining the memory address of the memory unit to be tested, the method further includes: Obtain the test task status of the memory stress test; When the test task status is the execution end status, the step of obtaining the memory address to be tested of the memory unit to be tested is performed.
8. The method according to claim 7, characterized in that The obtaining of the test task status of the memory stress test includes: When the execution of a single memory stress test algorithm ends or the execution of all the memory stress test algorithms ends, determining that the test task state is the execution end state; When the memory pressure test algorithm is not finished executing, the test task state is determined to be an executing state.
9. The method according to claim 1, characterized in that: The memory test process to be tested is obtained based on the memory address to be tested and a preset address mapping relationship; the preset address mapping relationship is used to represent the mapping relationship between the memory address and the memory stress test process.
10. A server, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory is used to store computer-executable instructions; The processor is used to execute the computer-executable instructions stored in the memory to implement the memory unit abnormality determination method as described in any one of claims 1-9.
